Nom du GPU | R250 | GA100 |
---|---|---|
Architecture | Rage 7 | Ampere |
Fonderie | TSMC | TSMC |
Taille du processus | 150 nm | 7 nm |
Transistors | 60 million | 54,200 million |
Taille des matrices | 68 mm² | 826 mm² |
Date de publication | Never Released | Jun 22nd, 2020 |
---|---|---|
Génération | Radeon R200 | Tesla |
Production | End-of-life | Active |
Bus interface | AGP 4x | PCIe 4.0 x16 |
Prédécesseur | Radeon R100 | — |
Successeur | Radeon R300 | — |
GPU clock | 300 MHz | 1410 MHz |
---|---|---|
Mémoire clock | 300 MHz 600 Mbps effective | 1215 MHz 2.4 Gbps effective |
Base clock | — | 765 MHz |
Boost clock | — | 1410 MHz |
Taille de la mémoire | 128 MB | 40 GB |
---|---|---|
Type de mémoire | DDR | HBM2E |
Bus mémoire | 128 bit | 5120 bit |
Bande passante | 9.600 GB/s | 1,555 GB/s |
Pixel shaders | 4 | — |
---|---|---|
Vertex shaders | 2 | — |
TMUs | 8 | 432 |
ROPs | 4 | 160 |
Shading units | — | 6912 |
Comptage SM | — | 108 |
Tensor cores | — | 432 |
Cache L1 | — | 192 KB (per SM) |
Cache L2 | — | 40 MB |
Taux de pixel | 1.200 GPixel/s | 225.6 GPixel/s |
---|---|---|
Vertex rate | 150.0 MVertices/s | — |
Taux de texture | 2.400 GTexel/s | 609.1 GTexel/s |
FP16 (half) performance | — | 77.97 TFLOPS (4:1) |
FP32 (float) performance | — | 19.49 TFLOPS |
FP64 (double) performance | — | 9.746 TFLOPS (1:2) |
Taille Slot | Single-slot | Dual-slot |
---|---|---|
TDP | 25 W | 250 W |
Suggestions pour le PSU | 200 W | 600 W |
Sorties | 1x DVI1x VGA1x S-Video | No outputs |
Connecteurs de puissance | None | 8-pin EPS |
Longueur | — | 267 mm 10.5 inches |
Board number | — | P1001 SKU 200 |
DirectX | 8.1 | — |
---|---|---|
OpenGL | 1.4 | — |
OpenCL | — | 3.0 |
Vulkan | — | — |
Pixel shader | 1.4 | — |
Vertex shader | 1.1 | — |
CUDA | — | 8.0 |
Shader model | — | — |